NOTICE: The design principles laid out in this document are now reality,
in the form of Expert Quake.
The Gameplay Guide is a resource for all of the various people who are
extending id software’s Quake: QuakeC programmers, level editors, everyone.
This Guide is about the one thing that I don’t see anyone talking about
- how the various add-ons that people have written alter the game play
and game flow of Quake.
- An Example - the Doomsday Device
- Skills, Focus and Design
- The Aspects
- Expert Quake
- Evaluating and Tinkering with Gameplay
- Map Design
- 1/9/97: published
- 1/12/97: new section “Player
Knowledge and Control over Outcome” under The Archetypes
- 1/18/97: new sections “deathmatch
3” and “Equalized Weapons“
under the simple changes section, products of an email conversation with
- 1/18/97: added references to new sections, changed “Quad
Damage and the Pentagram of Protection“
- 1/19/97: clarified the Guide’s purpose in the Introduction
- 1/19/97: almost completely rewrote and added to “Maps
- 1/19/97: new section “A Better
- 1/20/97: new section “Weird
- 2/07/97: inserted section “Games,
Skills, and the Game Designer“, stealing some content that used
to be under “The Archetypes“
- 2/11/97: added to section “Equalized
Weapons“, included Axe and Single Shotgun in patch file
- 2/19/97: added a table under “Maps
and Terrain“, made small clarifications throughout, added colors
- 2/26/97: added a link to the Balance
patch, new conclusion in the Doomsday Device
- 2/26/97: retitled “Memorization Skills” to
“Specific Knowledge as Skill“,
and rewrote it.
- 2/27/97: rewrote “Games,
Skills and the Game Designer“
- 3/2/97: adding a reference to a new, related page called
“The Peanut Gallery“
- 3/8/97: added to “Relative
Attack and Defense Strength..“, explaining the other extreme.
- 3/29/97: rewrote “Range
of Power and Item Fetching“, removed “Good Powerups and Trivial
- 3/30/97: rewrote “Goals,
Simulations and alternate game modes“
- 7/30/97: full rewrite after experience implementing Expert
Quake. Map design material moved to a separate
document. Some links in above log will no longer work.
How do you define good gameplay?
- It’s possible to increase in skill almost indefinitely - variety and
- Score is directly proportional to skill at the most difficult tasks
- balance and focus
These two criteria define a notion of good gameplay, which you’ll
not find in a dictionary. This notion of gameplay is not necessarily
directly equivalent to how fun a multiplayer game is.
Certainly, a chaotic multiplayer game in which freely available insanely
powerful weapons kill groups of other players in one shot can be a very
fun game, and a game I might play - for a while.
I tend to lose interest in any game where it’s not possible to keep
improving one’s ability, or where improvements in ability don’t show through
in the game very much. If this isn’t at least a little bit true of you,
then ignore this document, and go back to work on your “kill everyone
on the level by pressing a button” Quake mod (sorry, couldn’t resist).
But if you’re interested in creating a game with what I’ve defined as “good
gameplay”, or at least in creating a game that finds a balance
between gameplay and more chaotic carnage, read on.
The Guide is designed to give you a very strong and clear understanding
of gameplay, to the point that you are able to predict the effect on gameplay
that any arbitrary change to the game will have, and clearly explain why
it would have that effect. Once you can do that, you can make informed
decisions about whether to implement a given change or not, weighing the
effect of the change on gameplay against the extent to which the change
will make the game more engaging aside from its effect on gameplay.
Let’s say you have a Quake-like game with a Doomsday Device. The Doomsday
Device is the only weapon in the game, and there is only one on
the map, in the same place every time. Each player spawns and runs for
the device, uses it to kill all players (instantly) and score a bunch of
points. All players spawn again and run for the Doomsday Device. This happens
Every player quickly learns the location of the Doomsday Device, and
the best path to it from any of the various spawn points. Barring players
too incompetent to pick up on these simple things and navigate the level,
after a certain amount of time every player will have roughly equal score.
That is, the players are all perfect experts at this particular
game. This game conforms perfectly to criteria two - that skill matches
score. It fails criteria one - the amount of skill that can be acquired
is very limited.
Let’s say you take Quake and add the Doomsday Device. The majority
of the time, whoever was spawned closest to the Doomsday Device triggers
it and instantly kills all other players. Occasionally, two players who
weren’t going to be the first to the Doomsday Device would get into a toe-to-toe
fight that would entail all of the skills and complexity of Quake, and
the player who fought better might get a single frag before the Doomsday
Device went off. The skill of each player would be a fixed amount representing
the number of times they happened to be nearest to the Doomsday device,
plus some amount much smaller than this, that would be the number of times
they got a frag before the Doomsday Device went off. Now, to the extent
that it was possible to acquire skill in Quake, you can acquire skill in
Doomsday Quake. However, the Doomsday Device is still the best way to get
points, even though it takes much less skill to use the Doomsday Device
than to get a frag in combat.
At this point, you may be thinking, “so all I have to do is resist
the urge to add a Doomsday Device to Quake?”. Well, no. The Doomsday
Device is not really one particular modification that can ruin gameplay.
It is a representative of a whole class of possible game modifications
that have varying degrees of the same detrimental effect. The bottom
line on Doomsday Devices is that they create a way to score points without
requiring as much skill as other means of scoring points, which is
a gameplay problem that can occur with much less extreme and less easily
recognizable analogs of the Doomsday Device, such as a powerup that is
There are whole other classes of problems, with various extreme and
mundane examples of what kind of game modification can cause that class
of problem. This Guide is about exploring and identifying the problems,
the causes, and the solutions.
What is meant by “a skill”? A skill is the ability to take
the correct action in a given situation. Aim, the ability to hit where
you intend, is a skill. The ability to determine where to aim is another
skill, or set of skills. Combat is a set of skills.
By understanding and classing sets of skills, we can redefine or come
to a clearer definition of good gameplay, phrased in terms of the traits
of various skills.
A Continuum of skills: Execution vs Knowledge
What skills are involved in gathering items efficiently, in a free-for-all
deathmatch? Probably the most critical factor is familiarity with the level.
That’s a piece of knowledge - is it also a skill? By the definition above,
a skill is an ability. The ability to gather items efficiently is essentially
level knowledge plus the trivial ability to know your location in a level
- in a free-for-all situation you are unlikely to know anything else of
importance. Gathering items is a skill that is 80% knowledge and 20% execution.
Other skills, such as knowing where to be in combat, are much less reducible
to knowledge. You can say “stay away from walls or it barely matters
if a rocket misses you” but you cannot say “here is what you
do in combat: …”.
A skill that has been literally reduced to knowledge is clearly a skill
that cannot be further developed. If a game is to have depth and complexity,
skills that can be reduced to knowledge must not be important in the game.
Let’s dispense immediately with the two most common arguments about
“It’s not fair”
Fairness is when two players are playing the same game. The Doomsday
Device game is perfectly fair. Quake is perfectly fair. The only way a
game could be unfair is if you had a systematic enforced difference in
how the game behaved for different players, for instance, if one player
always started with half the health of another player. Even randomness
is fair, as long as there is time for the random effects to apply equally
to all players.
“You’re may be winning, but only because you do cheap things that
take no skill”
Everything takes skill. Pressing a key takes skill. Looking at your
monitor and recognizing some set of pixels as a “player” takes
Both of these arguments are really trying address the same central problem
in game design, they are just badly expressed. Some skills have seemingly
endless depth, and can never be truly mastered. Other skills reduce to
a simple trick that can be taught to a newbie.
If you can learn a set of simple tricks and be as good or better at
a game as someone who has great expertise in the truly deep skills, that
is a shallow game with poor gameplay. It is neither unfair nor does it
take no skill. It is just a shallow game that has bad gameplay.
In particular, to get good gameplay, mere presence of deep skills in
the game is not enough. The deep skills in the game must be the sole determiners
of the outcome of the game, and no other simpler skills can in any way
detract from the central role of deep skills.
The Deep Skills of Quake
The deepest skills in Quake are easily recognizable, and most players
readily agree on what they are.
In Quake deathmatch, and probably in any free-for-all first person shooter,
the deepest skills are involved in beating one or more other players in
an even combat that lasts several seconds. These are the skills
of anticipation, using and understanding the offensive and defensive importance
of terrain and positioning, and precision in aim and movement, to name
a few. Even combat, as opposed to heavily lopsided combat, allows otherwise
inconsequential differences in mastery of deep skills to have importance.
A duration of several seconds all skills other than reflexes to come into
In any teamplay mode, coordination and teamwork, both in combat and
in such things as planned attacks and holding key positions, are additional
The task, then, is to make sure that deep skills are the only skills
that have significance in the game. Other skills, such as item gathering,
should either be eliminated or the game should be altered such that they
have little or no importance. This is not accomplished by adding new things
to a game, rather, it is an engine-up redesign.
In this redesign, several core aspects of multiplayer games in general
can be called out and analyzed to determine what should be done to focus
on deep skills. These are non-specific properties such as pace, information
acquisition, power and “health” levels, and scoring.
Skills that reduce to specific knowledge must be eliminated or made
unimportant wherever possible, in order to allow deep skills to determine
In Quake, specific knowledge as a skill is a problem that comes up most
often when players don’t know a level well or at least equally well.
An example of this problem is when one player knows about a useful secret
on a map that another player does not. The fact of knowing or not knowing
about the secret is a “skill” that may have importance in the
game, possibly having as much importance as combat skill.
One solution to the problems of differing level knowledge is to exclusively
use well known and well circulated maps, such as the id levels or the levels
distributed for a particular mod. A more general solution can be had at
the level of map design, that is, maps can be
made so that there is little specific knowledge of importance, so that
the “skill” of knowing a particular level well is just maxed
out in all players, as early on as possible, and therefore becomes a non-factor,
allowing deeper skills to determine the score.
Alternately, for a one-time match up, a level that no players have
seen might be used. In this case, it’s no longer true that specific
knowledge of the level dominates the game, instead, the skill of quickly
learning the level becomes important. This is a complex and somewhat
developable skill, but also arguably a very mechanical skill, based on
The balance between the ability to do damage and the ability to take
damage is probably the most critical aspect of a game in terms of gameplay.
Simply put, the more powerful a weapon is relative to the amount of damage
a player can take, the more important a single hit is in the game.
The more important a single hit is in the game, the more important
making the first hit is. When single hits are very significant in
the game, fast target recognition and fast reaction time become the most
important skills in the game, lowering the importance of deep skills
like intentional use of terrain.
Furthermore, the first hit will not always go to the quickest player.
When the first hit is pivotal, how does a player guarantee that they get
the first hit (maybe even the only hit)? By ambushing, sniping,
and other such tactics, all of which involve more shallow skills, such
as the ability to recognize a good ambush position, and not choke on the
free shot. The core idea behind ambushing is, after all, finding an easier
way to kill someone than killing them in even combat.
Preventing ambushing, sniping, reflexes and ping from dominating a game
is accomplished by ensuring that the first hit doesn’t all but determine
the outcome of the ensuing combat. In Quake this can be achieved by through
lower weapon lethality.
Even when the first shot doesn’t all but determine the outcome of combat,
ambushing is still a somewhat effective tactic that has a bad effect
on gameplay. This is problem best solved in map design.
It’s also possible to make individual hits too unimportant. The primary
effect of this is stalemates between players of equal skill. Aside
from outright stopping the progress of the game and of the score, stalemates
tend to end in ways that don’t involve skill, such as succumbing to a trap
or eventual carelessness.
When finding the middle point between purely reflexive play and long
wars of attrition, the skill of the very best players should be kept in
mind. Good players may make much more effective use of weapons, to the
extent that the game again becomes based on reflexes for those good players.
If there is the possibility of gaining power during a game, the focus
of the game shifts away from the ability to win even combats, allocating
importance to whatever skills are involved in gaining and maintaining power.
In Quake free for all deathmatch, this means shallow skills like item gathering
and weapon and powerup guarding become strong components of the game, fundamentally
Item gathering, however, is not the only possible basis for a range
of power in a game. If the skill involved in gaining and maintaining power
is as deep as combat itself, at least a moderate range of power can be
implemented while maintaining good gameplay. In particular, winning
a combat is strong basis for gaining power.
Where possible, the high end of a range of power should involve powerful
defenses rather than powerful offenses, so that powerful players in the
game are just difficult to defeat, rather than being so powerful that they
negate their opponent’s ability to fight back, or even evade them.
Scoring is meant to be a measure of each player’s skill at the game.
Deep and shallow skills can only be emphasized or de-emphasized with respect
to the scoring system.
In Quake deathmatch, score is based on number of kills, yet players
can become orders of magnitude more powerful than each other, making it
extremely easy to score kills. The most difficult task in the game is still
beating another player who is at least as powerful as you, yet this
is not what is scored on, nor does it correspond very well to kills.
One possibility for repairing this problem is to invent a scoring system
that reflects the possesion of deep skill rather than straight kills. A
system of weights could be devised according to weapon and armor effectiveness,
that would reward points independent of actual kills. Of course, this would
create a surreal game in which players continually vied for the position
of underdog in any given combat, avoiding powerups and superior weapons.
However, I’ve only mentioned this possibility in order to emphasize
the role of scoring; a cleaner, easier solution is to create a game in
which combats are even as often as possible, so that scoring a kill is
an indication of superiority in deep skills. This is done through balanced
weapons and items, and through appropriate restoration systems.
Discrete Scoring and the role of Restoratives
If an even combat is the unit of scoring, what happens when a combat
may be uneven due to damage dealt in a previous combat? The usual answer
to this is restoratives - health packs, armors, etc.
One option open to the game designer is that no restoratives are made
available. In a one on one game where there is no range of power, this
will mean the score will roughly reflect the total amount of damage each
player has done to the other. However, in a free for all, the benefit of
damage previously done to opponents will be traded around randomly between
players as variously uneven combats are joined.
The role of restoratives in solving this problem is essentially to allow
players to return to a baseline health whenever they are not under fire.
Considering typical situations in Quake, in particular use of terrain as
barriers, or three and four way combats, it’s clear that there is a continuum
from “safety” to “under fire”. When restoration powerups
are plentiful and evenly distributed, they can best express this continuum
of danger, in the relative rates of losing health to enemies and gaining
health by covering ground.
A similar effect can be achieved with continuous regeneration, or several
other methods. The key result is always that combats are even as often
Skills, Randomness and Information Availability
If a skill is the ability to take the correct action in a situation,
then each and every skill is heavily dependant on the information available
to a player. In particular, from the perspective any individual player
trying to make a decision in the game, randomness is exactly the same
thing as any other lack of information, such as not being able to see
all players at all times. A lack of information or a random behavior in
the game may make it impossible for a player to take the “ideal”
action in a situation.
Initially, it may seem that making as much information available to
the player as is possible is the right approach, in order to give the player
the maximum ability to take correct actions. However, total information
can as easily destroy good gameplay as create it - imagine a game in
which you could see all opponents at all times, even through other solid
objects. Skills like anticipation of opponents not visible to you, or losing
players who are chasing you, would be gone. A game like Quake with total
information would likely reduce to head fakes at 90 degree corridor turns;
other games might cause players to experience information overload - there
would be a best action to take, but understanding what that action is would
involve an intractable number of factors.
Total information and a total lack of randomness is not necessary or
even wanted. Rather, it should simply be the case that good decisions
do not ever lead to bad results. That is, given all data available
to him, a skillful player should be able to take an action with a net benefit
in every situation.
In terms of design, find a way to make the unknowable either knowable
or unimportant. For instance, not being able to see all other players at
all times is fine, as long as it is possible to navigate a level such that
ambushes are avoidable and combats are entered on even terms. Similarly,
the fact that a player has a big powerup should be obvious, like the sounds
and glow that indicate the Quad Damage.
Among other applications of this rule, consider spawnfrags in Quake.
Spawnfrags represent a low, random probability of getting killed when moving
through some position. Even if a player has memorized all spawn positions
in a level, in a combat situation against an equally skilled opponent,
moving through a spawn position may represent a net reduction of risk -
the player makes a good decision that leads to being telefragged.
My ongoing answer to how to achieve good gameplay in Quake is the Expert
Quake modification. It is neither perfect nor complete, but has gone a
long, long way toward good gameplay. The source for Expert Quake is freely
available at the Expert Quake
home site, on the downloads
Rationalization and Doomsday Devices
One way to be blind to a problem in gameplay is to confuse the possible
in theory with the possible due to skill. If a player fires a Doomsday
Device into a room and obliterates a player who never saw or heard anything
that would have told him to dive into the corner, it really doesn’t
matter if he could have survived by diving into the corner. It wasn’t
possible to survive due to skill, and that’s the only thing that
matters for gameplay.
I’ve sometimes run into arguments such as, “the rocket launcher
should be the most powerful because it would be in real life”. This
is a bid for realism. My usual response is “Good point. You should
see this mod I made where it takes 45 seconds to put on armor and getting
shot with any gun kills or totally disables you, and..”.
Computer games are carved out of thin air. They can behave in any way
we want them to; modelling reality is one thing we can do. The strongest
reasons for realism are probably aesthetics and familiarity. Outside of
badly violating a player’s sense of aesthetics or reality, create whatever
kind of play you like. In my case that’s a focus on deep skills, but there
is plenty of room for simulations, games that are 90% determined by die
rolling, or by social interaction. Just be sure you’re clear on where you
stand when you criticize a design decision.
Fighting fire with more fire
It’s been explained to me many times that such and such super-powerup
is actually a good thing, because it can be used to kill players who have
gotten so many powerups that they are impossible to kill otherwise. This
is a matter of trying to solve the symptom instead of the problem that
brought it about - the player shouldn’t have been able to become so
powerful that players of at least close to the same skill level
find him impossible to kill. Introducing a super-wowie-zowie powerup
to kill super-powerful players just makes it so that players who are dominant
because of skill are also easily whacked, in addition to giving easy kills
to whoever happens to be wielding the super powerup at the time.
Signs of Excellent Gameplay
A key sign of excellent gameplay is that it’s possible to have two players,
both of whom are capable of easily beating an average player, where one
of these players is still capable of easily beating the other. This
indicates that a truly broad range of skills can exist, and also that when
differences in skills exist, they affect the score.
This is not the only criteria however. Certainly there are levels and
levels of ability at Quake that become quite visible, at least in one on
one matches. This is an indication that Quake contains deep skills,
and that the masters know the shallow skills too, so that matches between
masters can still reflect deep skill to an extent.
Gameplay and Fun
An emphasis on deep skill may bring to mind the image of a game for
only the most hardcore gamers, going at it on a level few can even comprehend.
The truth is that a game with good gameplay is an excellent learning environment
for new players - there is a lack of shallow skills and randomness clouding
the core of the game. The reasons for losing can be directly witnessed
and understood by watching the combat moves of the player who defeats you.
Players are much less likely to yell “cheater!” or blame deaths
on luck, ping, or whatever.
Still, good gameplay is not for every player. Light gamers who primarily
appreciate things like massive explosions will not be interested by a game
focusing on deep skills. Immature players may want a game where a super-powerup
allows a marginally better player to dominate a player who could otherwise
be his near-match. Simulation fanatics will want realism. Other players
will want yet other things, or a mix.
As for me, I like to learn.
Map design in the context of good gameplay is as large, important and
complex a topic as design of game behavior. Multiplayer
Map Design is a second section of the Designer’s Guide, discussing
map design aimed at good gameplay.
Thanks for reading the Gameplay Guide. I hope you found it more useful
Like anything on the web, the Gameplay Guide is perpetually a work in
progress. If you have something to add, or a question that you want to
ask, please feel free to email me.
Copyright 1997 Charles Kendrick, all rights reserved.